1. Field of the Invention
This invention is directed to recovering clock pulses of wavelength division multiplexed optical signals. In particular, it relates to simultaneous clock recovery of many wavelength division multiplexed optical signals.
2. Technical Background
As the capacity of wavelength division multiplexed (WDM) transmission systems increases in response to the increasing demand for communication, the maximum reach of each transmission system is diminished. Regenerators are therefore required at regular intervals along a transmission link in addition to any regenerators associated with network nodes where traffic routing takes place. It may be argued that regenerators are necessary within switching nodes to provide traffic routing and grooming functions, though this is not always the case when traffic on a given wavelength is routed straight through the node. However, the use of regenerators between nodes increases the network cost without contributing additional functionality. A cost-effective means of regenerating WDM signals is therefore required as an alternative to full WDM demultiplexing and opto-electronic regeneration. System manufacturers indicate that this is particularly necessary for 40 Gbit/s data rate systems with a target reach of 3000 km but a practical transmission limit around 1500 km.
A 3R regenerator (Reamplifying, Reshaping, Retiming) is a known example of an all-optical regenerator useful for future high-speed and high-capacity transparent optical networks. All-optical clock recovery is a major building block of the 3R all-optical regenerator because clock recovery is needed for its re-timing function. Many single channel approaches to all-optical clock recovery have been proposed and demonstrated. One single-channel clock recovery device used a fiber-optic parametric oscillator where the amplitude-modulated parametric gain for the clock signal is optically phase insensitive. Most clock recovery approaches are designed for one channel operation because for multi-channel all-optical clock recovery (MOCR), technical challenges are multiplied.
In a first MOCR approach, two-channel optical clock recovery was demonstrated using stimulated Brillouin scattering (SBS) in an optical fiber. However, due to the wavelength dependence of the Brillouin frequency shift, the total optical bandwidth effectively available to this clock recovery device is only about 3 nm. This limited spectral coverage is a severe drawback of the SBS-based MOCR. In a second approach, MOCR was achieved in an actively mode-locked fiber ring laser formed by a semiconductor optical amplifier array module integrated with two waveguide grating routers (AWGs) and an Er-doped fiber amplifier (EDFA). Several significant disadvantages exist with this approach. First, because of the homogeneous line broadening of the EDFA, the multi-channel operation of the fiber laser is inherently unstable. Second, in this device, each semiconductor optical amplifier (SOA) in the array module acts as an active mode-locker for only one corresponding channel. This increases the cost and complexity of the system. Third, no means to compensate the difference in path lengths for different channels within the SOA-AWG block were implemented, which is a requirement for multi-channel operation. Finally, overall speed of the device is still limited by the speed of the SOA response.
Therefore there is a need for an improved method and apparatus for use in all-optical clock recovery and signal regeneration, which can simultaneously process a plurality of WDM signals.